Modelling and optimization of arsenic removal from municipal water using Fe-Al deposited activated carbon derived from neem leaf waste: A response surface methodology approach

Abstract

The contamination of drinking water with arsenic is a serious ecological and social health problem, especially in impoverished areas and rural territories. The work aims at a low-cost and environmental friendly synthesis of an effective arsenic removal adsorbent using Fe-Al modified activated carbon prepared using Hawthorn Neem leaf waste material. The objective of the study is to model, optimize, and statistically confirm the process of adsorption through the analysis of such parameters as the adsorbent dose (1-5 g/L), the adsorption experiments were conducted using arsenic-contaminated water with influent concentrations ranging from 50 to 300 µg/L, representing levels commonly reported in affected groundwater systems rate (2-10 mL/min) as well as the contact time (30-180 min). The interactive effects and the determination of the best operating conditions are determined using a Response Surface Methodology (RSM) that runs on Central Composite Design (CCD). The developed models exhibited high statistical significance (p < 0.001, where p denotes the probability of obtaining the observed results by chance) and strong goodness of fit, as indicated by high coefficients of determination (R², which represents the proportion of variance in the experimental data explained by the model). Optimization outcomes demonstrate that within influent concentration of 100 µg/L, 3.5 g/L adsorbent dose and a flow rate of 5 mL/min, the ability of removal of arsenic is more than 98.2%. Several empirical and semi-empirical models commonly reported in the literature for describing fixed-bed adsorption behaviour such as the Thomas, Yoon–Nelson, and Adams–Bohart models were employed to analyze the column adsorption dynamics. Among these, the Thomas model showed the best agreement with the experimental breakthrough data (R² = 0.982), followed by the Yoon–Nelson (R² = 0.974) and Adams–Bohart (R² = 0.961) models.